Impact of AI-powered virtual triage and care referral on patient care seeking behavior in a Middle Eastern health plan

Abstract

Objective: To assess whether automated AI-based virtual triage and care referral (VTCR) improves appropriate acuity-based care by aligning patient healthcare seeking intent and potential care seeking behavior with triage output in a leading multinational healthcare plan based in the Middle East.

Methods: Data were derived from an AI-based symptom checker application and analyzed the pre- and post-VTCR care intentions of eligible health plan patients (N = 4,985) to examine how VTCR influenced potential care seeking behavior across five levels of care acuity. Pre- and post-triage care intentions were compared, and changes as a result of triage, including acuity level escalation and de-escalation, were assessed for statistical significance using Z-tests.

Results: Overall alignment with VTCR clinical guidance was 37.6% following virtual triage, improved from a 22.2% level of acuity alignment prior to VTCR. VTCR significantly decreased the number of patients with uncertain healthcare intention (62.9% or - 22.1 PP; p = .05), the largest group of whom decided to engage self-care after VTCR (13.9% of all patients). The largest changes in care intent occurred where patients altered their care plan to engage self-care (an increase of 128.2% or +18.8 PP; p = .05), reducing avoidable use of higher acuity services. Post-triage intent to access emergency care increased 138.8% (+ 1.9 PP; p = .05). The largest de-escalation of care acuity was observed among patients who before VTCR intended to engage a non-urgent outpatient consultation, but instead chose self-care after VTCR (9.3% of patients; p = .05).

Conclusions: Virtual triage reduced potential clinically inappropriate utilization of both higher and lower acuity care services by patients, and post-VTCR care seeking was better aligned with patients’ actual clinical needs. VTCR improved early detection of and care referral for emergent conditions, and simultaneously reduced inappropriate ED and outpatient care utilization for symptoms that could be managed by patients through self-care. VTCR was able to reduce care acuity-level misalignment and potentially unnecessary and avoidable healthcare utilization.